Associations of Distance to Trauma Care, Community Income, and Neighborhood Median Age With Rates of Injury Mortality | Emergency Medicine | JAMA Surgery | JAMA Network
[Skip to Navigation]
Sign In
Figure 1.  Study Sample Flowchart
Study Sample Flowchart

Composition of study sample and matching process. EMS, emergency medical services; MIEMSS, Maryland Institute for Emergency Medical Services Systems.

Figure 2.  Spatial Distribution of Zip Code Tabulation Area–Level Injury Incidence in Maryland
Spatial Distribution of Zip Code Tabulation Area–Level Injury Incidence in Maryland

Injury incidence by zip code tabulation area per 1000 adult residents in 2015.

Table 1.  Estimated Population Characteristics and Bivariate Mortality Association With Injury Incidents
Estimated Population Characteristics and Bivariate Mortality Association With Injury Incidents
Table 2.  Injury Incident Mortality in Maryland by Individual Characteristics of Affected Persons
Injury Incident Mortality in Maryland by Individual Characteristics of Affected Persons
Table 3.  Adjusted Associations of Injury Incidents With Mortality in Maryland by Characteristics of the Built and Social Environment
Adjusted Associations of Injury Incidents With Mortality in Maryland by Characteristics of the Built and Social Environment
1.
Centers for Disease Control and Prevention. Injury prevention & control data & statistics web-based injury statistics query and reporting system. http://www.cdc.gov/injury/wisqars/. Published August 1, 2017. Accessed August 4, 2017.
2.
Sorenson  SB.  Gender disparities in injury mortality: consistent, persistent, and larger than you’d think.  Am J Public Health. 2011;101(suppl 1):S353-S358.PubMedGoogle ScholarCrossref
3.
Lehmann  R, Beekley  A, Casey  L, Salim  A, Martin  M.  The impact of advanced age on trauma triage decisions and outcomes: a statewide analysis.  Am J Surg. 2009;197(5):571-574.PubMedGoogle ScholarCrossref
4.
Haider  AH, Weygandt  PL, Bentley  JM,  et al.  Disparities in trauma care and outcomes in the United States: a systematic review and meta-analysis.  J Trauma Acute Care Surg. 2013;74(5):1195-1205.PubMedGoogle Scholar
5.
Cubbin  C, LeClere  FB, Smith  GS.  Socioeconomic status and injury mortality: individual and neighbourhood determinants.  J Epidemiol Community Health. 2000;54(7):517-524.PubMedGoogle ScholarCrossref
6.
Downing  SR, Oyetunji  TA, Greene  WR,  et al.  The impact of insurance status on actuarial survival in hospitalized trauma patients: when do they die?  [published correction appears in J Trauma. 2011;70(2):525].  J Trauma. 2011;70(1):130-134.PubMedGoogle ScholarCrossref
7.
Tohira  H, Jacobs  I, Mountain  D, Gibson  N, Yeo  A.  Systematic review of predictive performance of injury severity scoring tools.  Scand J Trauma Resusc Emerg Med. 2012;20:63.PubMedGoogle ScholarCrossref
8.
Thompson  HJ, Rivara  FP, Nathens  A, Wang  J, Jurkovich  GJ, Mackenzie  EJ.  Development and validation of the mortality risk for trauma comorbidity index.  Ann Surg. 2010;252(2):370-375.PubMedGoogle ScholarCrossref
9.
Carr  BG, Bowman  AJ, Wolff  CS,  et al.  Disparities in access to trauma care in the United States: a population-based analysis.  Injury. 2017;48(2):332-338.PubMedGoogle ScholarCrossref
10.
Hsia  R, Shen  YC.  Possible geographical barriers to trauma center access for vulnerable patients in the United States: an analysis of urban and rural communities.  Arch Surg. 2011;146(1):46-52.Google ScholarCrossref
11.
Carr  BG, Branas  CC, Metlay  JP, Sullivan  AF, Camargo  CA  Jr.  Access to emergency care in the United States.  Ann Emerg Med. 2009;54(2):261-269.PubMedGoogle ScholarCrossref
12.
Mullins  RJ.  A historical perspective of trauma system development in the United States.  J Trauma. 1999;47(3)(suppl):S8-S14.PubMedGoogle ScholarCrossref
13.
American College of Surgeons Committee on Trauma.  Regional Trauma Systems: Optimal Elements, Integration, and Assessment. Chicago, IL: American College of Surgeons; 2008.
14.
Rogers  FB, Rittenhouse  KJ, Gross  BW.  The golden hour in trauma: dogma or medical folklore?  Injury. 2015;46(4):525-527.PubMedGoogle ScholarCrossref
15.
Brown  JB, Rosengart  MR, Billiar  TR, Peitzman  AB, Sperry  JL.  Distance matters: Effect of geographic trauma system resource organization on fatal motor vehicle collisions.  J Trauma Acute Care Surg. 2017;83(1):111-118.PubMedGoogle ScholarCrossref
16.
Glance  LG, Osler  TM, Mukamel  DB, Dick  AW.  Impact of trauma center designation on outcomes: is there a difference between Level I and Level II trauma centers?  J Am Coll Surg. 2012;215(3):372-378.PubMedGoogle ScholarCrossref
17.
MacKenzie  EJ, Rivara  FP, Jurkovich  GJ,  et al.  A national evaluation of the effect of trauma-center care on mortality.  N Engl J Med. 2006;354(4):366-378.PubMedGoogle ScholarCrossref
18.
Merchant  JK, Kross  B, Donham  KJ, Pratt  DS, eds.  Agriculture at Risk: A Report to the Nation. Kansas City, MO: National Coalition for Agricultural Safety and Health; 1989.
19.
Peek-Asa  C, Zwerling  C, Stallones  L.  Acute traumatic injuries in rural populations.  Am J Public Health. 2004;94(10):1689-1693.PubMedGoogle ScholarCrossref
20.
Winkleby  MA, Cubbin  C.  Influence of individual and neighbourhood socioeconomic status on mortality among black, Mexican-American, and white women and men in the United States.  J Epidemiol Community Health. 2003;57(6):444-452.PubMedGoogle ScholarCrossref
21.
Haider  AH, Hashmi  ZG, Zafar  SN,  et al.  Minority trauma patients tend to cluster at trauma centers with worse-than-expected mortality: can this phenomenon help explain racial disparities in trauma outcomes?  Ann Surg. 2013;258(4):572-579.PubMedGoogle Scholar
22.
Maryland Institute for Emergency Medical Services Systems. Maryland state trauma registry data dictionary for adult patients. https://www.miemss.org/home/Portals/0/Docs/OtherPDFs/Web-registry-data-dictionary-adult.pdf?ver=2016-03-10-140444-337. Published January 15, 2014. Accessed December 26, 2017.
23.
Maryland Geographic Information Office. Geographic information office (GIO) overview. http://imap.maryland.gov/Pages/gio-office.aspx. Accessed August 4, 2017.
24.
United States Census Bureau. American consumer survey technical documentation. https://www.census.gov/programs-surveys/acs/technical-documentation.html. Published October 13, 2016. Accessed August 4, 2017.
25.
Clark  DE, Osler  TM, Hahn  DR. ICDPIC: Stata module to provide methods for translating International Classification of Diseases (Ninth Revision) diagnosis codes into standard injury categories and/or scores. https://ideas.repec.org/c/boc/bocode/s457028.html. Published October 29, 2010. Accessed August 4, 2017.
26.
Benson  M, Koenig  KL, Schultz  CH.  Disaster triage: START, then SAVE—a new method of dynamic triage for victims of a catastrophic earthquake.  Prehosp Disaster Med. 1996;11(2):117-124.PubMedGoogle ScholarCrossref
27.
Charlson  ME, Pompei  P, Ales  KL, MacKenzie  CR.  A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.  J Chronic Dis. 1987;40(5):373-383.PubMedGoogle ScholarCrossref
28.
Jerrett  M, Gale  S, Kontgis  C.  Spatial modeling in environmental and public health research.  Int J Environ Res Public Health. 2010;7(4):1302-1329.PubMedGoogle ScholarCrossref
29.
Milcent  C.  Hospital ownership, reimbursement systems and mortality rates.  Health Econ. 2005;14(11):1151-1168.PubMedGoogle ScholarCrossref
30.
Alter  HJ, Braun  R, Zazzali  JL.  Health status disparities among public and private emergency department patients.  Acad Emerg Med. 1999;6(7):736-743.PubMedGoogle ScholarCrossref
31.
Burns  R, Nichols  LO, Graney  MJ, Applegate  WB.  Mortality in a public and a private hospital compared: the severity of antecedent disorders in Medicare patients.  Am J Public Health. 1993;83(7):966-971.PubMedGoogle ScholarCrossref
32.
Ashour  A, Cameron  P, Bernard  S, Fitzgerald  M, Smith  K, Walker  T.  Could bystander first-aid prevent trauma deaths at the scene of injury?  Emerg Med Australas. 2007;19(2):163-168.PubMedGoogle ScholarCrossref
33.
Shah  MN, Bazarian  JJ, Lerner  EB,  et al.  The epidemiology of emergency medical services use by older adults: an analysis of the National Hospital Ambulatory Medical Care Survey.  Acad Emerg Med. 2007;14(5):441-447.PubMedGoogle ScholarCrossref
34.
Berwick  DM, Downey  AS, Cornett  E, eds.  A National Trauma Care System: Integrating Military and Civilian Trauma Systems to Achieve Zero Preventable Deaths After Injury. Washington, DC: The National Academies Press; 2016.
35.
Maryland Institute for Emergency Medical Services Systems. 2013-2014 annual report. http://msa.maryland.gov/megafile/msa/speccol/sc5300/sc5339/000113/020000/020584/unrestricted/20141678e.pdf. Published 2015. Accessed December 26, 2017.
Original Investigation
June 2018

Associations of Distance to Trauma Care, Community Income, and Neighborhood Median Age With Rates of Injury Mortality

Author Affiliations
  • 1Center for Surgery and Public Health Department of Surgery, Brigham and Women’s Hospital, Boston, Massachusetts
  • 2Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, Johns Hopkins University, Baltimore, Maryland
  • 3Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Johns Hopkins University, Baltimore, Maryland
  • 4Department of Surgery, Johns Hopkins School of Medicine, Johns Hopkins University, Baltimore, Maryland
JAMA Surg. 2018;153(6):535-543. doi:10.1001/jamasurg.2017.6133
Key Points

Question  What is the association of injury incident location characteristics with mortality?

Findings  This cross-sectional study of 16 082 adults found an 8% increase in the odds of death for every 5-mile increase in distance to the nearest trauma center and an 3% increased odds of death for every 5-year increase in neighborhood median age. Odds of death decreased by 27% when neighborhood per capita income was greater than $25 000.

Meaning  Characteristics of injury incident locations may contribute to injury mortality and should be considered when planning trauma care systems and injury prevention efforts.

Abstract

Importance  Rural, low-income, and historically underrepresented minority communities face substantial barriers to trauma care and experience high injury incidence and mortality rates. Characteristics of injury incident locations may contribute to poor injury outcomes.

Objective  To examine the association of injury scene characteristics with injury mortality.

Design, Setting, and Participants  In this cross-sectional study, data from trauma center and emergency medical services provided by emergency medical services companies and designated trauma centers in the state of Maryland from January 1, 2015, to December 31, 2015, were geocoded by injury incident locations and linked with injury scene characteristics. Participants included adults who experienced traumatic injury in Maryland and were transported to a designated trauma center or died while in emergency medical services care at the incident scene or in transit.

Exposures  The primary exposures of interest were geographic characteristics of injury incident locations, including distance to the nearest trauma center, designation level and ownership status of the nearest trauma center, and land use, as well as community-level characteristics such as median age and per capita income.

Main Outcomes and Measures  Odds of death were estimated with multilevel logistic regression, controlling for individual demographic measures and measures of injury and health.

Results  Of the 16 082 patients included in this study, 8716 (52.4%) were white, and 5838 (36.3%) were African American. Most patients were male (10 582; 65.8%) and younger than 65 years (12 383; 77.0%). Odds of death increased by 8.0% for every 5-mile increase in distance to the nearest trauma center (OR, 1.08; 95% CI, 1.01-1.15; P = .03). Compared with privately owned level 1 or 2 centers, odds of death increased by 49.9% when the nearest trauma center was level 3 (OR, 1.50; 95% CI, 1.06-2.11; P = .02), and by 80.7% when the nearest trauma center was publicly owned (OR, 1.81; 95% CI, 1.39-2.34; P < .001). At the zip code tabulation area level, odds of death increased by 16.0% for every 5-year increase in median age (OR, 1.16; 95% CI, 1.03-1.30; P = .02), and decreased by 26.6% when the per capita income was greater than $25 000 (OR, 0.73; 95% CI, 0.54-0.99; P = .05).

Conclusions and Relevance  Injury scene characteristics are associated with injury mortality. Odds of death are highest for patients injured in communities with higher median age or lower per capita income and at locations farthest from level 1 or 2 trauma centers.

Introduction

Injuries are responsible for millions of life-years lost annually in the United States.1 Associations of age, sex, race/ethnicity, socioeconomic status, injury severity, and comorbidities with mortality are well documented.2-8 People living in rural, low-income, and historically underrepresented minority communities face substantial barriers to trauma care,9-11 which suggests that features of the health care system, built environment, and social environment may contribute to differences in treatment and outcomes. Unfortunately, evidence of the effects of the built and social environment on injury outcomes is limited.

Time to treatment is a major consideration in trauma system design12,13 and the primary indicator of environmental barriers to trauma care in the literature. Most studies14 indicate some benefit from shorter prehospital times, though the ideal treatment window varies. To our knowledge, only 1 study to date15 has examined trauma center proximity as an independent determinant of mortality, demonstrating increased motor vehicle crash fatality rates with increasing distance from the crash site to the trauma center.

Hospital characteristics also predict injury mortality, with improved outcomes for patients treated at level 1 and 2 trauma centers compared with hospitals with fewer resources.16,17 Because of the inherent association between injury incident location, trauma center location, and the course of injury treatment, it is possible that characteristics of the hospital nearest an injury incident scene determine mortality, even if the patient is treated at a different hospital.

In addition, injury incidence and severity may be elevated in rural settings owing to differences in land use patterns,18,19 but, to our knowledge, no studies have specifically examined land use as a determinant of injury outcomes. Community-level social measures of trauma center patient populations may be independent determinants of mortality,5,20,21 but it is not clear if the social environment at the injury scene has a similar impact on outcomes. The lack of evidence concerning the potential association of injury scene characteristics with mortality limits the impact of interventions that address modifiable features of the social environment that disproportionately affect vulnerable populations.

The Maryland trauma system is one of the oldest in the United States and is regarded as an ideal model of trauma system organization. Using individual injury incident data from the Maryland Institute for Emergency Medical Services Systems (MIEMSS), linked with measures of the built and social environment at injury incident scenes, this study examined incident location characteristics as determinants of mortality independent of individual patient demographics, health characteristics, and duration of the prehospital interval.

Methods
Data Sources

Injury incident records from the 2015 MIEMSS Adult Trauma Registry were linked with prehospital emergency medical services (EMS) data from the MIEMSS eMeds Patient Care Reporting System. Registry and eMeds data were collected per published MIEMSS standards.22 The most recent available land use data, which were from 2010, were retrieved from the Maryland Geographic Information Office.23 American Community Survey data for calendar years 2010 through 2014 were retrieved from the United States Census Bureau.24

This study was reviewed and approved by the Johns Hopkins Bloomberg School of Public Health Institutional Review Board. The study used secondary data without direct patient contact or intervention and was therefore conducted with a waiver of informed consent.

Population, Setting, and Data

Adults (18 years and older) who were injured in Maryland between January 1, 2015, and December 31, 2015; transported by a Maryland-based EMS company by ground ambulance or helicopter; treated at a designated trauma center; and who met MIEMSS inclusion criteria22 were included in this study. Patients transferred to a trauma center were included if all other criteria were met, as were patients who died in EMS care at the incident scene or in transit. Registry and prehospital records were linked using incident identification numbers and probabilistic algorithms. Registry records without matched EMS records were retained for analysis if injury scene location information was included in the registry record.

Measures

Injury mortality was the primary outcome of interest, with injuries considered fatal if the patient died at the injury scene, in transit, or at the hospital. Covariates included measures at the patient, location, and zip code tabulation area (ZCTA) level.

Age was categorized in 10-year increments based on dates of birth and age at injury incident. Race and ethnicity from medical records were examined owing to their established relationship with injury outcomes,4 and were combined in a single variable, measured as white, African American, Hispanic, or other.

Severity of injury was categorized based on Injury Severity Score (ISS) and Revised Trauma Score (RTS). We calculated ISS using the International Classification of Disease, Ninth Revision, Clinical Modification codes and the International Classification of Disease Programs for Injury Categorization Stata module,25 then categorized results as mild (with a score ≤9), moderate (10-15), severe (16-24), or critical (≥25). Based on prehospital triage criteria,26 unweighted RTS was categorized as mild (with a score of 12, the highest possible score), moderate (11), severe (4-10), or critical (≤3). Both the ISS and RTS categories were combined into a single measure of injury severity, with RTS used only when ISS was unavailable. Among the 7704 patients with both ISS and RTS available, 6539 (84.9%) were placed in the same category for both scores.

Using external cause of injury codes, the mechanism of injury was categorized as blunt, penetrating, blunt and penetrating, or other. Charlson Comorbidity Index scores were categorized as 0, 1 through 5, and 5 or greater, based on registry comorbidity codes and weights proposed by Charlson et al.27 Insurance status was coded as private, public, or no insurance, based on hospital payment records.

Prehospital time was calculated as the number of minutes elapsed from the initial 911 call to trauma center arrival. Owing to the characteristics of the MIEMSS dataset, treating hospitals were not identifiable in individual patient records, and actual distance traveled was not measured. To characterize the association between injury incident location and the overarching trauma system, trauma center distance was measured as the Euclidean distance between the incident scene and the nearest trauma center. All level 1 and 2 trauma centers in Maryland are in the Baltimore-Washington combined statistical area, minimizing differences in distance to care for patients bypassing the nearest trauma center. Hospital type of the nearest trauma center was coded as private level 1 or 2, public level 1 or 2, and private level 3 based on public records searches.

Land use was measured at the individual feature level and categorized as residential, commercial, industrial/agricultural, transport, institutional, or undeveloped. Median age in a given ZCTA was measured in years, and per capita income was categorized as greater than or less than $25 000 per person. Hospital transfer status was not included in regression models because of its collinearity with prehospital time.

Analytic Approach

We used ArcGIS, version 10.2.2 (Environmental Systems Research Institute) for all geocoding and measurement/attainment of spatial variables. Records were mapped using geocoded coordinates of the injury incident address. Records without exact addresses (n = 2198) were mapped using incident zip code centroids. Records were linked with ZCTA measures based on the ZCTA polygon within which the scene location or Zip code centroid fell. Records with exact addresses were linked with trauma centers using spatial joins based on the nearest feature, and with land use based on the features of the polygon each point fell within.

All other analyses were conducted using Stata, version 13 (StataCorp). Most missing data derived from missing location information; therefore, missing data were treated as missing at random and addressed with multiple imputation with predictive mean matching. A total of 10 imputations were used, with the 10 nearest neighbors included in the random selection for each iteration. Imputed variables included trauma center distance, hospital type, land use, insurance, patient age, severity, Charlson Comorbidity Index, per capita income per ZCTA, and median age per ZCTA.

Regression analyses were conducted using a randomly selected sample of 10 000 records. Simple logistic regression models were used to assess the unadjusted association between each measure and mortality. Multivariable analyses used multilevel logistic regression models with random intercepts for ZCTA. Scatterplots were examined to assess the need for interaction and spline terms. Associations of individual characteristics (model 1) and location characteristics (model 2) were modeled separately before combining all measures into a fully adjusted model (model 3). Regression analyses were considered statistically significant at α = .05. Standard spatial statistics methods were used to assess regression models for residual spatial dependence at the individual location (via a semivariogram of standardized residuals) and ZCTA level (with Moran I).28

Results
Population Characteristics

Figure 1 illustrates the composition of the study population and the matching process. The final analytic sample (n = 16 082) included 15 388 registry records and 727 prehospital deaths. Most registry records (13 157/16 082; 85.5%) were linked with a prehospital record, and an additional 2198 of 16 082 (14.3%) contained sufficient location information for inclusion. Thirty-three records (33/16 082; 0.2%) lacked location information and were excluded.

Figure 2 illustrates the spatial distribution of injury incidence per 1000 adults at the ZCTA level. Sample characteristics are presented in Table 1.

Age ranged from 18 to 100 years, with most patients younger than 65 years (77.0%). (Percentages reported for patient characteristics are estimated based on multiple imputation; denominators listed are based on the number of records with nonmissing data for each characteristic.) Most patients were male (10 581/16 082; 65.8%) and white (8346/14 311; 52.4%); most had mild injuries (14 166/15 242; 89.8%), blunt injuries (13 178/16 082; 81.9%), and no reported comorbidities (14 519/15 355; 94.6%). The proportions of patients with public insurance coverage (5518/13 698; 40.3%) and private insurance coverage (5560/13 698; 39.6%) were similar. The mean (SD) distance to the nearest trauma center was 9.9 (0.09) miles, and the mean (SD) total prehospital time was 64.6 (0.69) minutes. The nearest trauma center was private level 1 or 2 for 7076 incidents (44.0%) and public level 1 or 2 for 6770 incidents (42.1%). Residential land use (6609/16 082; 41.1%) and transportation land use (4953/16 082; 30.8%) were most common. The ZCTA median age, when averaged, was 38.2 years (SD, 0.04 years), and 11 949 (74.3%) of injury incidents occurred in a ZCTA with per capita income greater than $25 000.

Bivariate Analyses

Bivariate odds ratios (OR) and 95% CIs are presented in Table 1. Compared with incidents on residential land, the odds of death decreased by 38.6% for incidents on commercial land (OR, 0.61; 95% CI, 0.49-0.77; P < .001), by 47.6% for incidents on industrial/agricultural land (OR, 0.52; 95% CI, 0.37-0.75; P = .04), by 39.1% for incidents on transport land (OR, 0.61; 95% CI, 0.53-0.70; P < .001), and by 37.9% for incidents on institutional land (OR, 0.62; 95% CI, 0.46-0.83; P = .01). Odds of death increased by 1.7% per 5-minute increase in prehospital time (OR, 1.02; 95% CI, 1.01-1.03; P = .02).

Multivariable Analyses

Odds ratios and 95% CIs for all multivariable regression models are presented in Table 2 and Table 3. In model 1, patients with penetrating injuries were 3.88 times more likely to die than those with blunt injuries (OR, 3.88; 95% CI, 2.51-6.00 for penetrating injuries relative to blunt injuries, a reference category; P < .001), while patients with other injury types were 7.34 times more likely to die (OR, 7.34; 95% CI, 3.91-13.82; P < .001). For patients with penetrating injury, odds of death increased by 6.0% per 5-minute increase in prehospital time (OR, 1.06; 95% CI, 1.03-1.07; P < .001), while odds of death increased by 17.3% per 5-minute interval for patients with blunt and penetrating injuries (OR, 1.17; 95% CI, 1.14-1.18; P < .001) and by 4.8% per 5-minute interval for other injury mechanisms (OR, 1.05; 95% CI, 1.02-1.06; P < .001).

In model 2, odds of death increased by 11.1% per 5-mile increase in trauma center distance (OR, 1.11; 95% CI, 1.06-1.16; P < .001). Compared with incidents on residential land, incidents on commercial land were associated with a 35.8% decrease in odds of death (OR, 0.64; 95% CI, 0.49-0.84; P < .001), incidents on industrial/agricultural land showed a 52.2% reduction in odds of death (OR, 0.48; 95% CI, 0.30-0.76; P < .001), incidents on transportation land had 42.3% reduced odds of death (OR, 0.53; 95% CI, 0.35-0.81; P < .001), and incidents on institutional land had 47.3% reduced odds of death (OR, 0.53; 95% CI, 0.35-0.81; P < .001).

In model 3, patients with penetrating injuries were 5.28 times more likely to die than those with blunt injuries (OR, 5.28; 95% CI, 3.27-8.50 for penetrating injuries relative to blunt injuries, a reference category; P < .001), and patients with other types of injury were 8.26 times more likely to die (OR, 8.26; 95% CI, 4.32-15.77; P < .001). Odds of death increased by 5.7% per 5-minute increase in prehospital time for patients with penetrating injuries (OR, 1.06; 95% CI, 1.03-1.07; P < .001), by 15.4% per 5-minute increment for patients with blunt and penetrating injuries (OR, 1.15; 95% CI, 1.12-1.16; P = .01), and by 4.8% per 5-minute increment for patients with other injuries (OR, 1.05; 95% CI, 1.01-1.09; P = .01). Odds of death increased by 8.0% per 5-mile increase in trauma center distance (OR, 1.08; 95% CI, 1.01-1.15; P = .03). Compared with private level 1 or 2 centers, odds of death increased by 80.7% when the nearest trauma center was public level 1 or 2 (OR, 1.81; 95% CI, 1.39-2.34; P < .001), and increased by 49.9% when the closest trauma center was private and level 3 (OR, 1.50; 95% CI, 1.06-2.11; P = .02). Compared with incidents on residential land, odds of death decreased by 32.6% for incidents on commercial land (OR, 0.63; 95% CI, 0.49-0.92; P = .01), and increased by 70.0% for incidents on transportation land (OR, 1.70; 95% CI, 1.31-2.20; P < .001). Odds of death increased by 16.0% per 5-year increase in ZCTA median age (OR, 1.16; 95% CI, 1.03-1.30; P = .02). Per capita income greater than $25 000 was associated with a 26.6% decrease in odds of death (OR, 0.73; 95% CI, 0.54-0.99; P = .05), compared with per capita incomes below this threshold.

Discussion

To our knowledge, this study is the first to combine patient-level trauma center data and incident location information for an entire state of the United States, associating individual patient measures with characteristics of the built environment and social environment, regardless of injury mechanism, payment type, or hospital affiliation. To our knowledge, this is the first study to examine the associations of a broad range of spatially defined characteristics present at the injury scene with mortality while controlling for prehospital time and individual patient characteristics. Prior studies examined trauma center distance and designation as measures of trauma care accessibility,9-11,16,17 but treated distance as a proxy for prehospital time, not as an independent predictor of mortality.

Increasing distance from the injury scene to the nearest trauma center was associated with increased mortality independent of prehospital time, which suggests that the association between distance and mortality extends beyond distance as a determinant of time. This is consistent with recent findings from Brown et al15 of an association between motor vehicle fatality rates and trauma center distance. While injured patients are not always treated at the nearest trauma center, trauma center proximity reflects the accessibility of trauma care at a given location. Geographic variation in EMS service levels may contribute to the effect of distance on mortality, and the sequence and duration of prehospital events may vary by trauma center distance.

Level 1 or 2 designation of the trauma center nearest the injury scene was found to reduce mortality, which was consistent with prior evidence of benefits from treatment at level 1 or 2 centers.17 Private ownership of the nearest trauma center was associated with improved outcomes. Research on the association of hospital ownership with mortality is limited, but it does suggest a protective effect of private ownership, possibly due to increased resources, earlier adoption of technology,29,30 or protective health and socioeconomic characteristics among communities surrounding private hospitals.31

The results of this study suggest that the odds of death are lowest for injury incidents in commercial areas and highest in transportation areas. These findings conflict with prior hypotheses that agricultural and industrial land use spaces carry greater risk of mortality based on presumed differences in injury mechanism and severity.18,19 It is possible the public nature of commercial spaces confers some benefit following an injury incident owing to shorter EMS response times.32 In contrast, patients injured at transportation locations (eg, interstate highways) may experience prolonged prehospital times owing to incident-induced needs for traffic and/or vehicle extrication. The association between transportation land use and mortality changed drastically after adjustment for individual characteristics, suggesting populations injured at transportation locations are at low risk of mortality compared with other injury patients, potentially owing to injury characteristics associated with motor vehicle crashes.

The median age and per capita income per ZCTA were both associated with mortality. Odds of death were higher for incidents in older communities, possibly owing to the increased demands that an older population places on an EMS system.33 Per capita income less than $25 000 was also associated with increased odds of mortality, potentially reflecting the effect of individual socioeconomic indicators that were not measured in the registry, differences in the level of prehospital care in low-income communities, or differences in trauma centers serving predominantly populations of lower socioeconomic status. The association of ZCTA income levels with mortality is consistent with evidence that population level socioeconomic measures determine individual health outcomes.5,20,21

The observed association of African American race was inconsistent with prior studies,4 possibly because of the geographic distribution of African American residents in Maryland relative to trauma center locations. Prior studies of race as a determinate of injury mortality have not controlled for prehospital time or trauma center proximity.4 There is evidence that prehospital time is greater for African American patients than for white patients nationally10; however, shorter prehospital times were observed for African American patients in this study, which may contribute to better outcomes than expected. Given the association between prehospital time and mortality, time may mediate the association between race and mortality, warranting further examination.

Limitations

This analysis was limited to incident origin and transfer patients treated at designated trauma centers and did not include patients receiving definitive care at healthcare facilities other than trauma centers or those who died prior to EMS arrival at the injury scene. These patients represent a small but meaningful subset of patients facing substantial barriers to trauma care. This is a well-documented limitation of trauma systems research.34 However, comprehensive triage and transport protocols used by Maryland EMS professionals, an extensive air ambulance network, and the small geographic area of the state greatly reduce the number of high-risk injury patients treated at nontrauma centers,35 mitigating selection bias concerns.

Many records lacked information regarding the exact location of the injury incident and could not be linked with measures of the built environment. Patterns of missingness were examined by county, mortality outcome, demographic characteristics, and month of incident to rule out likely causes of nonignorable nonresponses. Data appeared to be missing at random and were imputed using predicted mean matching. Sensitivity analyses indicated little impact from missingness, but it is possible an unknown or unmeasured covariate was causally associated with missing location information.

Finally, generalizability of this study is limited by the unique organization of the Maryland trauma care system34 and the small size of the state. In many ways, the Maryland system represents a best-case scenario for the delivery of trauma care, with clear, standardized prehospital protocols implemented throughout the state. States with more variation in care may see greater effects of geographic barriers. Analyses conducted in this study should be replicated with data from other states that represent a range of approaches to EMS and trauma system organization.

Conclusions

This study demonstrates the association of several environmental measures with mortality while also confirming previously demonstrated associations of individual demographic and injury characteristics. Distance from the injury scene to the nearest trauma center is a strong geographic determinant of injury mortality. Trauma center distance may be associated with differences in timing, sequence, or quality of prehospital care. Observed associations of community-level median age and per capita income suggest that the level and quality of EMS services in a community depends on EMS resources and use of services. Future studies should examine the sequence and duration of events during the prehospital interval, paying attention to changes in the prehospital experience associated with trauma center proximity, community demographics, and EMS professional characteristics.

Back to top
Article Information

Corresponding Author: Molly Jarman, PhD, Center for Surgery and Public Health, Department of Surgery, Brigham and Women’s Hospital, 1620 Tremont St, 4-020, Boston, MA 02120 (mjarman@bwh.harvard.edu).

Accepted for Publication: November 23, 2017.

Published Online: February 7, 2018. doi:10.1001/jamasurg.2017.6133

Author Contributions: Dr Jarman had full access to all data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Jarman, Castillo, Curriero.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Jarman.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Jarman, Curriero, Castillo.

Obtained funding: Jarman.

Administrative, technical, or material support: Jarman, Castillo.

Study supervision: Haut, Pollack Porter, Castillo.

Conflict of Interest Disclosures: Drs Pollack Porter and Castillo are both core faculty with the Johns Hopkins Center for Injury Research and Policy. Dr Haut received royalties from Lippincott, Williams, Wilkins for the book Avoiding Common ICU Errors and received payment as the author of a paper commissioned by the National Academies of Medicine titled “Military Trauma Care’s Learning Health System: The Importance of Data Driven Decision Making,” which was used to support the report titled, “A National Trauma Care System: Integrating Military and Civilian Trauma Systems to Achieve Zero Preventable Deaths After Injury.” No other disclosures were reported.

Funding/Support: Dr Jarman’s effort was supported by Agency for Healthcare Research and Quality National Research Services Award grant T32HS000029 and by a William Haddon Jr fellowship from the Johns Hopkins Center for Injury Research and Policy.

Role of the Funder/Sponsor: The Agency for Healthcare Research and Quality had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Drs Castillo and Pollack Porter are affiliated with the Johns Hopkins Center for Injury Research and Policy; other than the author contributions outlined above, the Center had no additional role in the design and conduct of the study; collect, management, analysis and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

References
1.
Centers for Disease Control and Prevention. Injury prevention & control data & statistics web-based injury statistics query and reporting system. http://www.cdc.gov/injury/wisqars/. Published August 1, 2017. Accessed August 4, 2017.
2.
Sorenson  SB.  Gender disparities in injury mortality: consistent, persistent, and larger than you’d think.  Am J Public Health. 2011;101(suppl 1):S353-S358.PubMedGoogle ScholarCrossref
3.
Lehmann  R, Beekley  A, Casey  L, Salim  A, Martin  M.  The impact of advanced age on trauma triage decisions and outcomes: a statewide analysis.  Am J Surg. 2009;197(5):571-574.PubMedGoogle ScholarCrossref
4.
Haider  AH, Weygandt  PL, Bentley  JM,  et al.  Disparities in trauma care and outcomes in the United States: a systematic review and meta-analysis.  J Trauma Acute Care Surg. 2013;74(5):1195-1205.PubMedGoogle Scholar
5.
Cubbin  C, LeClere  FB, Smith  GS.  Socioeconomic status and injury mortality: individual and neighbourhood determinants.  J Epidemiol Community Health. 2000;54(7):517-524.PubMedGoogle ScholarCrossref
6.
Downing  SR, Oyetunji  TA, Greene  WR,  et al.  The impact of insurance status on actuarial survival in hospitalized trauma patients: when do they die?  [published correction appears in J Trauma. 2011;70(2):525].  J Trauma. 2011;70(1):130-134.PubMedGoogle ScholarCrossref
7.
Tohira  H, Jacobs  I, Mountain  D, Gibson  N, Yeo  A.  Systematic review of predictive performance of injury severity scoring tools.  Scand J Trauma Resusc Emerg Med. 2012;20:63.PubMedGoogle ScholarCrossref
8.
Thompson  HJ, Rivara  FP, Nathens  A, Wang  J, Jurkovich  GJ, Mackenzie  EJ.  Development and validation of the mortality risk for trauma comorbidity index.  Ann Surg. 2010;252(2):370-375.PubMedGoogle ScholarCrossref
9.
Carr  BG, Bowman  AJ, Wolff  CS,  et al.  Disparities in access to trauma care in the United States: a population-based analysis.  Injury. 2017;48(2):332-338.PubMedGoogle ScholarCrossref
10.
Hsia  R, Shen  YC.  Possible geographical barriers to trauma center access for vulnerable patients in the United States: an analysis of urban and rural communities.  Arch Surg. 2011;146(1):46-52.Google ScholarCrossref
11.
Carr  BG, Branas  CC, Metlay  JP, Sullivan  AF, Camargo  CA  Jr.  Access to emergency care in the United States.  Ann Emerg Med. 2009;54(2):261-269.PubMedGoogle ScholarCrossref
12.
Mullins  RJ.  A historical perspective of trauma system development in the United States.  J Trauma. 1999;47(3)(suppl):S8-S14.PubMedGoogle ScholarCrossref
13.
American College of Surgeons Committee on Trauma.  Regional Trauma Systems: Optimal Elements, Integration, and Assessment. Chicago, IL: American College of Surgeons; 2008.
14.
Rogers  FB, Rittenhouse  KJ, Gross  BW.  The golden hour in trauma: dogma or medical folklore?  Injury. 2015;46(4):525-527.PubMedGoogle ScholarCrossref
15.
Brown  JB, Rosengart  MR, Billiar  TR, Peitzman  AB, Sperry  JL.  Distance matters: Effect of geographic trauma system resource organization on fatal motor vehicle collisions.  J Trauma Acute Care Surg. 2017;83(1):111-118.PubMedGoogle ScholarCrossref
16.
Glance  LG, Osler  TM, Mukamel  DB, Dick  AW.  Impact of trauma center designation on outcomes: is there a difference between Level I and Level II trauma centers?  J Am Coll Surg. 2012;215(3):372-378.PubMedGoogle ScholarCrossref
17.
MacKenzie  EJ, Rivara  FP, Jurkovich  GJ,  et al.  A national evaluation of the effect of trauma-center care on mortality.  N Engl J Med. 2006;354(4):366-378.PubMedGoogle ScholarCrossref
18.
Merchant  JK, Kross  B, Donham  KJ, Pratt  DS, eds.  Agriculture at Risk: A Report to the Nation. Kansas City, MO: National Coalition for Agricultural Safety and Health; 1989.
19.
Peek-Asa  C, Zwerling  C, Stallones  L.  Acute traumatic injuries in rural populations.  Am J Public Health. 2004;94(10):1689-1693.PubMedGoogle ScholarCrossref
20.
Winkleby  MA, Cubbin  C.  Influence of individual and neighbourhood socioeconomic status on mortality among black, Mexican-American, and white women and men in the United States.  J Epidemiol Community Health. 2003;57(6):444-452.PubMedGoogle ScholarCrossref
21.
Haider  AH, Hashmi  ZG, Zafar  SN,  et al.  Minority trauma patients tend to cluster at trauma centers with worse-than-expected mortality: can this phenomenon help explain racial disparities in trauma outcomes?  Ann Surg. 2013;258(4):572-579.PubMedGoogle Scholar
22.
Maryland Institute for Emergency Medical Services Systems. Maryland state trauma registry data dictionary for adult patients. https://www.miemss.org/home/Portals/0/Docs/OtherPDFs/Web-registry-data-dictionary-adult.pdf?ver=2016-03-10-140444-337. Published January 15, 2014. Accessed December 26, 2017.
23.
Maryland Geographic Information Office. Geographic information office (GIO) overview. http://imap.maryland.gov/Pages/gio-office.aspx. Accessed August 4, 2017.
24.
United States Census Bureau. American consumer survey technical documentation. https://www.census.gov/programs-surveys/acs/technical-documentation.html. Published October 13, 2016. Accessed August 4, 2017.
25.
Clark  DE, Osler  TM, Hahn  DR. ICDPIC: Stata module to provide methods for translating International Classification of Diseases (Ninth Revision) diagnosis codes into standard injury categories and/or scores. https://ideas.repec.org/c/boc/bocode/s457028.html. Published October 29, 2010. Accessed August 4, 2017.
26.
Benson  M, Koenig  KL, Schultz  CH.  Disaster triage: START, then SAVE—a new method of dynamic triage for victims of a catastrophic earthquake.  Prehosp Disaster Med. 1996;11(2):117-124.PubMedGoogle ScholarCrossref
27.
Charlson  ME, Pompei  P, Ales  KL, MacKenzie  CR.  A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.  J Chronic Dis. 1987;40(5):373-383.PubMedGoogle ScholarCrossref
28.
Jerrett  M, Gale  S, Kontgis  C.  Spatial modeling in environmental and public health research.  Int J Environ Res Public Health. 2010;7(4):1302-1329.PubMedGoogle ScholarCrossref
29.
Milcent  C.  Hospital ownership, reimbursement systems and mortality rates.  Health Econ. 2005;14(11):1151-1168.PubMedGoogle ScholarCrossref
30.
Alter  HJ, Braun  R, Zazzali  JL.  Health status disparities among public and private emergency department patients.  Acad Emerg Med. 1999;6(7):736-743.PubMedGoogle ScholarCrossref
31.
Burns  R, Nichols  LO, Graney  MJ, Applegate  WB.  Mortality in a public and a private hospital compared: the severity of antecedent disorders in Medicare patients.  Am J Public Health. 1993;83(7):966-971.PubMedGoogle ScholarCrossref
32.
Ashour  A, Cameron  P, Bernard  S, Fitzgerald  M, Smith  K, Walker  T.  Could bystander first-aid prevent trauma deaths at the scene of injury?  Emerg Med Australas. 2007;19(2):163-168.PubMedGoogle ScholarCrossref
33.
Shah  MN, Bazarian  JJ, Lerner  EB,  et al.  The epidemiology of emergency medical services use by older adults: an analysis of the National Hospital Ambulatory Medical Care Survey.  Acad Emerg Med. 2007;14(5):441-447.PubMedGoogle ScholarCrossref
34.
Berwick  DM, Downey  AS, Cornett  E, eds.  A National Trauma Care System: Integrating Military and Civilian Trauma Systems to Achieve Zero Preventable Deaths After Injury. Washington, DC: The National Academies Press; 2016.
35.
Maryland Institute for Emergency Medical Services Systems. 2013-2014 annual report. http://msa.maryland.gov/megafile/msa/speccol/sc5300/sc5339/000113/020000/020584/unrestricted/20141678e.pdf. Published 2015. Accessed December 26, 2017.
×